Specialists from Sechenov University have developed the only machine learning model in Russia to predict USE scores. On its basis, a program has been created that can be useful to both teachers and students.
“The neural network analyzes the various achievements of the student,” explained Aleksey Ovsyannikov, head teacher of the Pre-University of Sechenov University for Innovative Development. – Achievements include grades in the subject, the results of olympiads and trial exams. The latest data is about 40% of the accuracy of neural network calculations. During the learning process, the neural network was taught to link these indicators and the scores that children received on real exams. Over the course of two years, the model was trained on hundreds of students.”
The work of the program was tested on several hundred students from the pre-university. The accuracy of her predictions was 85%, according to the developers. At the moment, the model has been worked out to predict the results of the exam in chemistry.
The prediction of USE results is complicated by poorly predictable human reactions under stress. Therefore, the further development of the model will be the inclusion of the temperament and psychotype of the student in it. This will take into account the impact of stress on the result and improve the accuracy of the forecast. The developers intend to analyze the influence of temperament on the assessment this year after passing the exam.
Read more in the exclusive Izvestia article:
Training alarm: the neural network will predict the result of passing the exam
Source: IZ
Jane Stock is a technology author, who has written for 24 Hours World. She writes about the latest in technology news and trends, and is always on the lookout for new and innovative ways to improve his audience’s experience.
